从电子健康记录中提取健康的社会决定因素:基于规则和基于大语言模型的方法的开发和比较
Bo Wang1,2,3,4, Dia Kabir1,2,3, Cheryl R Clark5
1Center for Precision Psychiatry, Massachusetts General Hospital, Boston, MA, United States.
大型语言模型 (LLM) 和基于规则的系统有效地从临床笔记中提取健康的社会决定因素 (SDoH). 结合这些方法为识别SDoH提供了一个可扩展的解决方案,改善了健康结果研究.
科学领域:
- 医疗信息学 医疗信息学
- 自然语言处理自然语言处理.
- 人口健康管理 人口健康管理
背景情况:
- 健康的社会决定因素 (SDoH) 显著影响健康结果,但在电子健康记录中通常记录不良.
- 准确的SDoH数据对于人口健康研究和临床信息学至关重要.
研究的目的:
- 开发和评估从非结构化的临床笔记中提取SDoH的可扩展方法.
- 将基于规则的系统 (RBS) 与用于SDoH提取的大型语言模型 (LLM) 进行比较.
- 评估组合模型的性能,将RBS和LLMs结合起来.
主要方法:
- 从171名患者的临床笔记创建了一个金级标准的SDoH体.
- 开发了一个基于规则的系统 (RBS),并与七个OpenAI GPT模型 (GPT-4o,4.1,4.1-mini,o4-mini,GPT-5,GPT-5-mini,o3) 相比较.
- 评估了零射击,少数射击设置和整体模型 (RBS-GPT晚期融合).
主要成果:
- 对于SDoH领域,RBS实现了高精度 (0.97),但回忆率低 (0.62).
- GPT模型,特别是GPT-5和GPT-5-mini (几次拍摄),在F1得分 (0.88域级) 中表现优于RBS.
- RBS-GPT组合模型在域级SDoH提取中达到0.89 F1分,精度和回忆均为平衡.
结论:
- 最先进的LLM,包括"迷你"GPT模型,在没有微调的情况下进行SDoH提取时表现出强大的性能.
- 整合RBS和LLMs可以提高SDoH识别的准确性和效率.
- 这一框架使得从临床文本中进行可扩展,具有成本效益的SDoH提取,用于研究和应用.
更多相关视频
05:47Evidence-based Knowledge Synthesis and Hypothesis Validation: Navigating Biomedical Knowledge Bases via Explainable AI and Agentic Systems
Published on: June 13, 2025
07:31Implementation of a Real-Time Psychosis Risk Detection and Alerting System Based on Electronic Health Records using CogStack
Published on: May 15, 2020
相关概念视频
Statistical Methods for Analyzing Epidemiological Data
Steps in Outbreak Investigation
Methods of Documentation VII: EMR
Classification of Illness
An illness is a response to a disease in which the person's level of functioning is changed compared with a previous level. The general classification of illness includes acute and chronic.
Acute illness is severe...
Models of Health Promotion and Illness Prevention I
The health belief model (HBM) attempts to predict health-related behavior in specific belief patterns. According to the HBM, a person's...
Issues And Trends In Healthcare Delivery System
Cost Containment
Payment for healthcare services has historically promoted adoption of costly and often unnecessary or inefficient...
